Article content
The field of industrial AI is steaming hot, with new algorithms and software coming out daily. Yet, the threshold for integrating AI into production may still be too high for small and medium-sized enterprises. This is where AI & Robotics Estonia (AIRE) comes to help.
Combining a business-oriented mindset with academic know-how
AI and robotics have huge potential to improve productivity and quality while reducing resource input. However, it requires several steps, from finding the right approach to testing before technological improvements take hold. Innovation is complicated, even for large enterprises. Let alone SMEs who need tailor-made solutions for their novel products and lack the resources for in-house development.
“We operate in the areas we see market failure,” explains Kirke Maar, head of AIRE. “We do not compete with existing solutions, but we aim to test more or less crazy ideas where we identify a need for scientific input.”
Operating as a structural unit of Tallinn University of Technology, AIRE combines the business-oriented mindset with the intellectual capacity of the academia. The University of Tartu and the University of Life Sciences are also part of AIRE.
“About half of our activities are awareness raising, educating, and auditing. The first step is to help our clients to understand where are they located at the roadmap towards digitalisation, and where are they heading,” says Ms Maar. “So, for example, in our educational program for engineers, CEOs and experts, we have already educated more than 300 people.”
Although the proximity to academia may lead to thinking of lecture halls with professors, AIRE aims to orient most of its training in a business-to-business approach.
“We like to have entrepreneurs who have experienced implementation of AI as speakers because they are credible and can speak the language of entrepreneurs,” says Ms Maar.
A centre for agile development
Another half of AIRE’s activities pertain to test before investing in demo projects in collaboration with industrial clients. Although AIRE has operated only for 2 years, it has had 144 clients so far, with a couple of new inquiries coming in each week. So far, clients have been from the industry sector, but AIRE aims to expand to the medical sector starting in 2024.
Probably part of the attractiveness of AIRE is that instead of long and large-scale R&D projects, their projects are small and fast. This approach is new for both entrepreneurs and academics, who are often used to a slower pace.
“We want to test new ideas and find out whether it works quickly,” says Ms Maar. “And we don’t hide it when it doesn’t, so we also publish and promote test cases where the approach failed. We would be too blue-eyed if we expected everything to work.”
This kind of fast iterative innovation is a good example of an agile mindset. It started in software development but is now implemented in other realms of innovation. AIRE is noteworthy because it aims for the public good to the largest extent. In addition to educating and publishing its test before investing in demo project results, it stores the output of software and algorithms on its own GitHub.
“Let’s be honest, it is often not the piece of code itself, but the skills and resources of implementing it and integrating it with the company’s production system that really matters,” adds Ms Maar.
Ambition to raise productivity by 25%
The overall aim of AIRE is not to implement AI and robotics for their own sake but to harness them for productivity growth. Ms Maar states that their declared ambition is to raise the digital maturity rate 25% of the segment’s average during the project period. After their projects have proven the workability of an idea or approach, they would move on to accelerators or science parks to develop them further and scale them up.
The portfolio of AIRE has examples from many walks of industrial production. Its projects have included companies that are aiming to get their first robot operational and ones with fully mechanised production plants that aim to operationalize their workflow.
Raiku, who plans to produce 100% compostable, beautiful protective packaging made of wood spirals, is an example of a fresh startup with global potential. They identified the need to harness the potential of AI and automation early on, and, lacking resources themselves, turned to AIRE.
“When producing packaging materials, we are talking about 10s of millions of spirals that are produced with 0.2 mm accuracy at tremendous speeds, and 24/7. No human is capable of quality controlling this,” says Karl Pärtel, co-founder and CEO.
With AIRE they were exploring the potential of machine vision for quality control, that would automatically streamline production also in case of defects.
“We are very satisfied with the process so far, as we have mapped and modelled the process and are currently teaching the AI so that we can start optimizing our production line in a month or so,” says Mr Pärtel.
Raiku is a good example of how harnessing AI has the potential to destabilize a whole industrial segment of fossils-based product – we are talking about bubble wrap here.
Thinking ahead with the AI
According to Ms Maar, AIREs test before invest demo-projects are mostly data-based, looking at how to harness AI better. Sometimes they start with only the potential of gathering data based on the AI strategy of the company and think about what could be done based on the data we don’t yet. The target is always on productivity, not AI itself as a technology.
The latter was the approach in the project with Valdek AS, a sheet-metal producer. According to Juhan Ernits, the head of the demo project, the goal of the project was to develop an artificial intelligence system using surveillance cameras that can monitor products stored on pallets with an accuracy of 2-3 meters, identify products and associate them with real labels.
“The project’s success is strongly related to the warehouse staff, who play an important role in teaching the system. It is for this reason that in the future we can see opportunities to implement the solution on a wider scale – it is enough to have great employees and cameras in the room, with which the system can see and learn,” says Mr Ernits.
Sometimes, the products and software have been implemented previously, but testing and developing are required to make a crucial component work better. A good example is Yanu, a robot waiter. The robot was already there, actively mixing and serving drinks to customers, but the stiffness of its movements hampered its operating speed. The goal of the demo project was to apply state-of-the-art algorithms to make the work of the service robot smoother and faster. So, the team filmed actual movements and assigned the robot to use AI to study and replicate them.
AIRE has already made a dent in the Estonian industry sector during its first years. As AIRE is part of the European Digital Innovation Hubs network. They are looking to create collaborations with hubs and businesses abroad and continue to contribute to the expansion of AI and robotics among small and medium-sized businesses.
